Diffuse between Two Species () and Two Species ()

Jing Zhang., Shuxia Wang., Houhun Li*., Bingbing Hu, Xiaofei Yang, Zhibo Wang College of Life Sciences, Nankai University, Tianjin, China

Abstract The diffuse coevolution between two species (Epicephala lativalvaris and E. mirivalvata) and two species (Breynia fruticosa and B. rostrata) is reported based on field observations and indoor experiments conducted in Hainan and Fujian, China. Study results showed that the two Epicephala species jointly pollinated the two Breynia species, which led to a unique obligate pollination mutualism of two2to2two species specificity. A single Epicephala larva exclusively fed on seeds of host and developed to maturity by consuming all six seeds of each fruit, whereas a fraction of intact fruits were left to ensure the reproduction of plants within the whole population. Larvae of the two Epicephala species are competitive for resources; the population of E. mirivalvata is much smaller than that of E. lativalvaris, which has resulted from the differences in the female ovipositor structures and oviposition mode. The life history of Epicephala species highly coincides with the phenology of Breynia plants, and different phenology of B. fruticosa resulted in the different life history of the two Epicephala species in Hainan and Fujian. The natural hybridization of two host plants, possibly induced by the alternate pollination of two Epicephala species, is briefly discussed.

Citation: Zhang J, Wang S, Li H, Hu B, Yang X, et al. (2012) Diffuse Coevolution between Two Epicephala Species (Gracillariidae) and Two Breynia Species (Phyllanthaceae). PLoS ONE 7(7): e41657. doi:10.1371/journal.pone.0041657 Editor: Dmitry A. Filatov, University of Oxford, United Kingdom Received February 14, 2012; Accepted June 25, 2012; Published July 27, 2012 Copyright: ß 2012 Zhang et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This research was supported by the National Natural Science Foundation of China (No. 30930014). Funder’s website: http://www.nsfc.gov.cn/. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected] . These authors contributed equally to this work.

Introduction carrying numerous pollen grains on proboscis, which led to a deduction that this species was an active pollinator of B. fruticosa The obligate pollination relationship between plants and their [11]. Okamoto et al. [12] suggested the floral scent of is seed-parasitic pollinators is perhaps one of the most specialized one of the important key signals that mediate the encounters of the cases in mutualism between and plants [1–4]. Ehrlich and species-specific partners in the Glochidion–Epicephala mutualism. In Raven defined coevolution as the interaction between two major tightly coevolved mutualism between Breynia vitis-idaea and groups of organisms with a close and evident ecological relation- Epicephala species, system-specific chemistry is not necessary for ship, such as plants and herbivores [5]. In the known classical efficient host location to Epicephala [13]. obligate pollination mutualism between figs–fig wasps and – By means of field observations and indoor experiments, we moths, figs and yuccas depend exclusively on pollination of studied the phenology, the phylogenetic relationship and the fig wasps and yucca moths respectively, while sacrificing some morphological characters of the two Breynia species, as well as the seeds for larvae of fig wasps and yucca moths to feed [6,7]. The morphological characters, the behavior, the life history, the obligate pollination mutualism between Epicephala moths of phylogenetic relationship and the population size of the two Gracillariidae and Glochidion plants of Euphorbiaceae was discov- Epicephala species in Hainan and Fujian. Based on related ered in 2003 [3]. Three pollination relationships were proposed literature and results of our study, we discussed herein: (1) the [8]: (1) one Epicephala species obligately pollinates one Glochidion impact of the different phenology of the allopatric B. fruticosa species, (2) two Epicephala species jointly pollinate one Glochidion populations on the life history of the Epicephala moths in species, and (3) two Epicephala species jointly pollinate two closely coevolution; (2) the influence of moth pollinators on the related parapatric Glochidion species. The most distinctive floral hybridization of Breynia species; and (3) whether the relationship features associated with the pollination mode are the structures of between the pollinators is stable in this diffuse coevolution. the pistils and stamens: styles are usually reduced to entire tips and medially fused, or filaments and anthers are variously fused [9– Results 10]. An obligate pollination relationship between Breynia vitis-idaea Life History of Epicephala Moths and Phenology of and an unnamed Epicephala species was discovered in 2004 [11]. Breynia Plants This Epicephala species actively pollinated B. vitis-idaea and was In Hainan, B. fruticosa has six fruit periods per year, lasting from observed loading numerous pollen grains on proboscis. Another the first period in early January to the sixth period in late Epicephala species on leaves of B. fruticosa was also observed November. Epicephala lativalvaris and E. mirivalvata have six

PLoS ONE | www.plosone.org 1 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Figure 1. Female flowers and fruits with double characters of Breynia plants. Female flowers (A) and fruits (B) with erect stigmas and flat calyx sepals. Female flowers (C) and fruits (D) with excurved stigmas and reflexed calyx sepals. (cs) calyx sepals. (s) stigma. doi:10.1371/journal.pone.0041657.g001 generations correspondingly, lasting from the first generation in A single larva of both E. lativalvaris and E. mirivalvata needs to middle January to the sixth generation in late December (Table consume all six seeds within a fruit to develop into maturity. S1). Mature larvae (the last instar) gnawed a hole in the fruit wall and In Fujian, both B. fruticosa and B. rostrata have 4–5 fruit periods exited, then produced cocoons and turned into pupae on leaves of per year, which lasted from the first period in early June to the fifth hosts or the ambient plants. Pupal stage of E. lativalvaris lasted 9– period in late April of the following year. A large number of 15 days, and pupal stage of E. mirivalvata lasted 9–12 days. Adults flowers withered in winter due to absence of pollinators. A small of both Epicephala species could survive 3–5 days by sucking nectar amount of flowers that were pollinated during middle to late for nutrition (Figure S1). November were dormant and no seeds were formed until middle or late March of the following year, and the fifth fruit period lasted Hybridization of Breynia Species from middle to late April. Both E. lativalvaris and E. mirivalvata have We found some plant individuals are difficult to identify, 4–5 generations correspondingly, which lasted from the first because they possess double morphological characters of both B. generation in middle June to the fifth generation in middle May of fruticosa and B. rostrata: three individuals (Figure 1: A, B) in Hainan the following year. Larvae that pupated during early to middle with erect stigmas (characteristic of B. rostrata) and flat calyx sepals November could emerge into adults, pollinate mature female (characteristic of B. fruticosa); nineteen individuals (Figure 1: C, D) flowers and lay eggs of the fifth generation during middle to late in Fujian with excurved stigmas (characteristic of B. fruticosa) and November. Larvae that pupated in late November overwintered in reflexed calyx sepals (characteristic of B. rostrata). pupas, and they began to emerge in late April and disappeared in The result of the hand-pollination hybridization experiment middle May. Pollinated flowers and Epicephala eggs in flowers were showed that there are no reproductive barriers between B. fruticosa dormant until middle or late March of the following year, and the and B. rostrata. 90% B. fruticosa female flowers (n = 50) with pollen adults of the fifth generation emerged in middle May (Table S2). grains from B. rostrata developed, so did 94% B. rostrata female We smelled flower fragrance of both B. fruticosa and B. rostrata at flowers (n = 50) with pollen grains from B. fruticosa (Table S7). night only, which suggests that they were blooming at night.

PLoS ONE | www.plosone.org 2 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Figure 2. Behavior of female Epicephala moths. Epicephala lativalvaris collecting pollen grains on male flower of Breynia fruticosa (A) and actively pollinating for Breynia fruticosa (B) and B. rostrata (C). E. lativalvaris inserting ovipositor through calyx lobe and ovary to lay eggs on B. fruticosa (D) and B. rostrata (E). (F) Egg of E. lativalvaris. E. mirivalvata laying eggs between ovary and calyx lobe on B. fruticosa (G) and B. rostrata (H). (I) Egg of E. mirivalvata. (e) egg. doi:10.1371/journal.pone.0041657.g002

Behavior of Epicephala Adults calyx lobes and ovary to lay eggs into ovary tissue (Figure 2: D–F). Epicephala moths became active after dark, which was in Epicephala mirivalvata females bent abdomen to basal calyx on dorsal accordance with the blossom of B. fruticosa and B. rostrata flowers. surface to deposit eggs between ovary and calyx lobes (Figure 2: Female adults of the two Epicephala species actively collected pollen G–I). After oviposition, Epicephala moths often gave complemen- from male flowers and pollinated female flowers of both Breynia tary pollination to make sure there are enough pollen grains for species at night. Female Epicephala adults pushed proboscis tip into female flowers. On rare occasion, they would not lay eggs if the dehiscent pit of calyx lobes, rubbed against stamen column to disturbed. collect pollen grains, and sometimes they pushed the whole proboscis into the pit (Figure 2: A). During pollen collection, Number of Epicephala Eggs and Consumption of Breynia female adults intermittently spread proboscis several times to get Seeds more pollen grains. They only collected pollen grains in one male Epicephala lativalvaris laid eggs by piercing through calyx lobes flower each time effectively, which could pollinate more than ten and ovary wall, leaving oviposition scars on fruits and calyx lobes. female flowers in the same or different plants. We could acquire their egg amount by counting scars on each Epicephala adults repeatedly rubbed proboscis against stigma to fruit. In all sampled fruits (n = 1269 in B. fruticosa, n = 867 in B. deposit pollen grains on all stigmas (Figure 2: B, C). They would rostrata), 72.0% in B. fruticosa (n = 914) and 88.6% in B. rostrata pace back and forth to pollinate all flowers on one branch, and (n = 768) with oviposition scars loaded E. lativalvaris eggs (Table female flowers were pollinated in this way as to pollinate most S3). Epicephala mirivalvata laid eggs between calyx lobes and ovary efficiently. wall. Therefore their eggs could only be observed by dissecting The two Epicephala species exhibited similar behavior of pollen female flowers. Only 2.6% female flowers in B. fruticosa (n = 2) and collection and pollination, but their oviposition behavior was 4.0% in B. rostrata (n = 5) loaded E. mirivalvata eggs in all dissected visibly different. Epicephala lativalvaris females bent abdomen to flowers (n = 76 in B. fruticosa, n = 126 in B. rostrata) (Table S4). We basal calyx on ventral surface and inserted ovipositor through

PLoS ONE | www.plosone.org 3 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Figure 3. Maximum-parsimonious (left) and maximum-likelihood (right) trees based on matK genes of 22 plant species. (A) The maximum-parsimonious tree of Phyllantheae plants (length, 179; CI = 0.933; RI = 0.985). (B) The maximum-likelihood tree of Phyllantheae plants (-ln likelihood = 2922.8917; transition/transversion ratio = 2; empirical frequencies: A = 0.30688, C = 0.16491, G = 0.15206, T = 0.37615). Numbers above branches are bootstrap values. Red branches refer to the species involved in this study. doi:10.1371/journal.pone.0041657.g003 collected 28 E. lativalvaris adults and two E. mirivalvata adults in the Discussion field, and reared 109 E. lativalvaris larvae and four E. mirivalvata larvae from fruits. We observed E. lativalvaris visiting female flowers Diffuse Coevolution between Two Epicephala Species 35 times in the field, but E. mirivalvata twice only. The above and Two Breynia Species statistical figures indicated that the population size of E. mirivalvata Diffuse coevolution firstly advanced and defined by Janzen was much smaller than that of E. lativalvaris. (1980) [14]. He defined coevolution as an evolutionary change in a trait of the individuals in one population in response to a trait of Phylogenetic Analysis of Epicephala Moths and Breynia the individuals of a second population, followed by an evolution- Plants ary response by the second population to the change in the first. We analyzed the COI genes of E. lativalvaris and Epicephala sp. ex Diffuse coevolution occurs when either or both populations in B. fruticosa [11], and found that the pairwise distance between them coevolution are represented by an array of populations that is 0. Moreover, Epicephala sp. ex B. fruticosa would leave scars on generate a selective pressure as a group [14]. host flowers after ovipositing, which is similar to E. lativalvaris. Our study revealed that B. fruticosa and B. rostrata distributed in Based on the above information, we could conclude that the Hainan and Fujian were dependent on the obligate pollination of Epicephala species pollinating B. fruticosa in Kawakita and Kato [11] E. lativalvaris and E. mirivalvata, forming a mutualism of is E. lativalvaris. two2to2two species specificity. Two Epicephala species deposited The topological structure of the maximum-parsimonious (MP) eggs on female flowers before or after pollination, and their larvae tree and the maximum-likelihood (ML) tree of plants are exclusively fed on seeds of two Breynia species. The life history of completely consistent (Figure 3), which indicates that B. fruticosa two Epicephala species correlated with the phenology of two Breynia and B. rostrata are highly surpported to be sister taxa. Both MP and species; Breynia plants bloomed after dark, which coincided with ML trees of Epicephala moths (Figure 4) showed that within five the activity of Epicephala moths. According to the definition of Epicephala pollinators of Breynia plants, E. mirivalvata diverged firstly diffuse coevolution and the classic examples of figs–fig wasps and from E. lativalvaris, Epicephala sp. ex B. vitis-idaea, Epicephala sp. ex B. yuccas–yucca moths, we confirm that diffuse coevolution exists disticha and Epicephala sp. ex B. oblongifolia, and that all of pollinators between the two Epicephala species and the two Breynia species, Epicephala of Breynia have a close relationship with Epicephala sp. ex P. vulcani. which is firstly revealed in this study. Though a single The pollinators of Glochidion plants and three non-pollinators of larva needs to consume all six seeds of each fruit to develop to maturity, enough intact fruits developed by Epicephala pollinated form a paraphyletic group. flowers are left to ensure the reproduction of the hosts within the whole population.

PLoS ONE | www.plosone.org 4 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Figure 4. Maximum-parsimonious (left) and maximum-likelihood (right) trees based on COI genes of 19 Epicephala moths species. (A) The maximum-parsimonious tree of Epicephala moths (length, 486; CI = 0.519; RI = 0.518). (B) The maximum-likelihood tree of Epicephala moths (- ln likelihood = 1628.20777; transition/transversion ratio = 2; empirical frequencies: A = 0.29923, C = 0.16930, G = 0.14101, T = 0.39046). Numbers above branches are bootstrap values. Red branches refer to the species involved in this study. doi:10.1371/journal.pone.0041657.g004

The impact of the phenology of the allopatric B. fruticosa on the fruticosa and B. rostrata are sister taxa (Figure 3). Breynia fruticosa and life history of Epicephala moths was obvious. Different climates in B. rostrata distributed in Wanshi Botanical Garden and Tianzhu Hainan and Fujian led to different phenology of B. fruticosa, which Mountain have similar phenology. When Epicephala moths resulted in different life history of Epicephala species. In Hainan B. collected pollen grains from male flowers of B. fruticosa, they are fruticosa has six fruit periods without dormant stage, so do two likely to deposit pollen grains on female flowers of the sympatric B. Epicephala species (Table S1). While in Fujian, pollinated female rostrata, hence resulting in the hybridization of B. fruticosa and B. flowers of B. fruticosa were dormant in winter, so a minority of rostrata, and vice versa. As B. fruticosa and B. rostrata are closely Epicephala individuals overwintered in eggs in female flowers and related, their hybrid offsprings could survive and possess double a majority overwintered in pupas until hosts bloom in the characters of both species. This hypothesis is supported by the fact following year (Table S2). that some plant individuals in Tianzhu Mountain and Wanshi Breynia fruticosa and B. rostrata are sister taxa, which suggests that Botanical Garden have double characters of both B. fruticosa and B. the pollinators they are sharing might have pollinated their rostrata (Figure 1). Breynia rostrata was recorded to have distribution ancestor, and that this pollination system maintained after the in Hainan [16], but we did not find it in this study. We only found speciation of B. fruticosa and B. rostrata. three individuals with characters of both B. rostrata and B. fruticosa in Yingge Mountain, which indicated that the population of B. Hybridization of Two Breynia Plants Induced by rostrata was either very small or has disappeared. The plants with Epicephala Moths double characters in Hainan may imply that B. rostrata once existed The study of the genetic structure of closely related figs and their there and might have hybridized with B. fruticosa in nature. associated pollinators in the Neotropical Region showed that shared use of host figs and colonization of novel hosts by the wasps Relationship between E. Lativalvaris and E. Mirivalvata may result in hybridization and genetic introgression across In Breynia–Epicephala diffuse coevolution mutualism, Epicephala different fig species [15]. We found similar phenomena in Breynia lativalvaris and E. mirivalvata overlap in the ecological niche, and are plants. Among the four species of Breynia plants in China, B. competitive for resources. Epicephala mirivalvata has a much smaller fruticosa and B. rostrata are closely related in morphology. Stamens population size than E. lativalvaris does and plays a secondary role concealed in closing calyx distinguish them from B. retusa, and free in this mutualism. This is closely associated with female ovipositor calyx and styles distinguish them from B. vitis-idaea. The structures and oviposition mode. Epicephala lativalvaris has an phylogenetic analysis based on matK gene also revealed that B. apically spine-shaped ovipositor (Figure 5: A), which could pierce

PLoS ONE | www.plosone.org 5 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Figure 5. Morphological characters of Epicephala female genitalia. (A) E. lativalvaris with spine-shaped apex. (B) E. mirivalvata with blunt apex. (o) ovipositor. (a) antrum. (db) ductus bursae. (cb) corpus bursae. doi:10.1371/journal.pone.0041657.g005 through calyx lobes and ovary wall to lay eggs into the ovary tissue population size, and E. lativalvaris may become the sole pollinator (Figure 2: D–F). So the external influence imposed on egg of B. fruticosa and B. rostrata in the future evolutionary course. hatching could be greatly reduced, and the newly hatched larvae could instantly feed on seeds to ensure a higher survival rate. On Conclusion the contrary, Epicephala mirivalvata has a clavate ovipositor with In Breynia–Epicephala diffuse coevolution, two morphologically blunt apex (Figure 5: B), which could deposit eggs only between and behaviorally distinct Epicephala species jointly pollinated two calyx lobes and ovary (Figure 2: G–I). So both eggs and newly Breynia species, forming a new obligate pollination mutualism of hatched larvae were exposed to the influence of environmental two–to–two species speciality. We firstly describe this model of factors. Moreover, the newly hatched larvae had to bore into the mutualism system in this study. Different phenology of B. fruticosa ovary, which could increase the mortal hazard and eventually resulted in different life history of Epicephala species in Hainan and decrease the population size. The phylogenetic analysis of Fujian, and two hosts sharing pollination of the two Epicephala Epicephala trees reveals that E. mirivalvata diverged firstly within species has led to hybridization of Breynia plants. In this mutualism, all known pollinators of Breynia plants, that E. mirivalvata firstly two competitive Epicephala species overlap in the ecological niche. established relationship with B. fruticosa and B. rostrata. However, its Epicephala mirivalvata is much smaller than E. lativalvaris in ovipositior structure and oviposition mode have resulted its population size due to different oviposition mode, and therefore inferiority in competition with E. lativalvaris. So it can be deduced faces more selection pressure. that E. mirivalvata is likely to disappear or shift host to maintain

PLoS ONE | www.plosone.org 6 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Figure 6. Flowers and fruits of Breynia plants. Male flowers of B. fruticosa (A) and B. rostrata (B) with stamens concealed in calyxe which can be visited by Epicephala moths only. Female flowers (C) and fruit (D) of B. fruticosa with excurved stigmas and discal calyx sepals. Female flowers (E) and fruit (F) of B. rostrata with stigmas erect and reflexed calyx sepals. (cs) calyx sepals. (s) stigma. doi:10.1371/journal.pone.0041657.g006

Materials and Methods Reserves (alt. 200–755 m, 18u509–19u129N, 109u159–109u509E), Wuzhi Mountain National Nature Reserves (alt. 512–720 m, Study Sites 18u499–18u599N, 109u329–109u439E) and Jianfeng Mountain From November 2009 to September 2011, we carried out field National Nature Reserves (alt. 807–973 m, 18u239–18u529N, observations on Breynia fruticosa in Yingge Mountain Nature 108u449–109u029E) in Hainan, China; and on B. fruticosa and B.

PLoS ONE | www.plosone.org 7 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia rostrata in Wanshi Botanical Garden (alt. 28–201 m, 117u539– Mountain, Wuzhi Mountain and Jianfeng Mountain) and 974 118u259E, 24u259–24u549N) and Tianzhu Mountain National fruits in Fujian (Tianzhu Mountain and Wanshi Botanical Garden) Forest Park (alt. 50–200 m, 24u359–24u399N, 117u539–117u579E) to acquire the egg amount of Epicephala moths (Table S4). in Fujian, China. Adult specimens examined in this study were collected from their hosts or reared from fruits in captivity. Dissection of female Studied Insects and Plants flowers of Breynia plants and female genitalia of Epicephala moths Epicephala lativalvaris Li, Wang & Zhang, 2012 and E. mirivalvata was conducted under an Olympus SZ11 stereo-microscope. Li, Wang & Zhang, 2012 are two small nocturnal moths that are Photographs were taken with a Canon G10 digital camera in distinctly different in morphology [17]. Females have a different the field and the illustration of genitalia were prepared with an oviposition mode due to having a differently shaped ovipositor Olympus C–7070 Wide Zoom digital camera. (Figure 5). However, they obligately and jointly pollinated B. We extracted genomic DNA of Epicephala moths from the fruits fruticosa in Hainan and Fujian, and pollinated B. rostrata in Fujian. we reared, and extracted genomic DNA of B. rostrata from silica- Breynia fruticosa and B. rostrata are similar in morphology: male gel dried leaves collected in Tianzhu Mountain, Fujian. The flowers (Figure 6: A, B) have a funnel-shaped calyx with six method of DNA extraction followed the CTAB procedure [19]. dentations at apex; the three stamens fused into a column and We PCR-amplified the COI gene fragments using primers LepF1 concealed in calyx, which makes it hard to be touched by flower and LepR1 [20], and PCR-amplified the matK gene fragments visitors except Epicephala moths; female flowers bear free calyx and using primers 570F and 1710R [21]. styles, which is likely to be accessed by other flower visitors. The We analyzed the phylogenetic relationship between E. lativalvaris two Breynia species can be distinguished by having different calyx and E. mirivalvata based on the mitochondrial cytochrome oxidase and stigmas: in B. fruticosa (Figure 6: C), calyx sepals radially spread subunit I (COI) gene of 20 Epicephala sequences, which is widely into a disk, stigmas are excurved and calyx sepals are enlarged in used in the phylogenetic analysis of insects [9,22,23]. The fruit (Figure 6: D); in B. rostrata (Figure 6: E), calyx sepals are sequences of E. lativalvaris, E. mirivalvata and E. relictella were newly reflexed, stigmas are erect and calyx sepals are not enlarged in fruit obtained during this study and 17 sequences were adopted from (Figure 6: F). the previous studies [8–9] (Table S5). Alignment of COI gene and matK gene were aligned using Mega v. 5.05. The phylogeny was Methods rooted with Epicephala sp. ex suffruticosa according to the Breynia fruticosa flowered for the 10 warmer months of the year, phylogenetic analysis of Epicephala and Phyllantheae [10]. fruited sparsely and irregularly in Hong Kong [18]. In order to We analyzed the phylogenetic relationship between B. fruticosa acquire detailed phenological data of B. fruticosa and B. rostrata as to and B. rostrata based on the maturase K (matK) gene, which has compare with the life history of E. mirivalvata and E. lativalvaris,we previously been shown to be particularly useful for phylogenetic made a tracking observation of the flower and fruit developmental analysis of the family and tribe [8,9,21,24]. Breynia rostrata sequence stages of B. fruticosa in Hainan, as well as of B. fruticosa and B. was newly obtained during this study and the remaining sequences rostrata in Fujian. We randomly selected ten individuals of B. were adopted from the previous studies [8,9,21] (Table S6). The fruticosa in Yingge Mountain Nature Reserves, and five individuals phylogeny was rooted with F. suffruticosa according to the of both B. fruticosa and B. rostrata in Tianzhu Mountain National phylogenetic analysis of Epicephala and Phyllantheae [10]. Forest Park and Wanshi Botanical Garden, respectively. We Maximum-parsimony and maximum-likelihood analyses of COI recorded the development of flowers and fruits, and counted the and matK were performed using PAUP v. 4.0 b10 [25]. Both number of flowers, the fruits and developing fruits of the selected maximum-parsimony and maximum-likelihood heuristic searches plant individuals once a week. were conducted with equal weight, 1000 and 100 replicates of To make sure whether the hybridization could happen between random addition analyses respectively, and tree bisection-recon- B. fruticosa and B. rostrata, we performed a hand-pollination nection branch swapping. The robustness of maximum-parsimony hybridization experiment in Wanshi Botanical Garden. We and maximum-likelihood trees were assessed by non-parametric respectively selected 50 non-pollinating female flowers of B. bootstrapping with 1000 and 100 replicates respectively. fruticosa and B. rostrata. We gathered pollen grains from male flowers of B. fruticosa and hand-pollinated female flowers of B. rostrata, and from male flowers of B. rostrata and hand-pollinated Supporting Information female flowers of B. fruticosa. We bagged these pollinated flowers Figure S1 Epicephala moths sucking nectar on female flowers of with fine netting (0.5 mm mesh), and seven days later, we counted Breynia fruticosa (A) and B. rostrata (B). and calculated their developmental rate (Figure S2). (TIF) We observed Epicephala moths during full anthesis and recorded their visiting behavior in detail, with particular attention paid to Figure S2 Hand-pollination hybridization experiment of Breynia their nocturnal activities. We recorded the time that the moths fruticosa and B. rostrata. (A) flowers bagged with fine netting. (B) spent on pollination, oviposition and pollen collection, and hand-pollinated B. rostrata with pollen of B. fruticosa. (C) non- observed how they used their proboscis to collect pollen and pollinated female flower of B. fruticosa. (D) developed female pollinate flowers, and where they oviposited. We collected flowers of B. fruticosa with pollen of B. rostrata. (E) non-pollinated developing and mature fruits and put them in a cylindrical plastic female flower of B. rostrata. (F) developed female flowers of B. box (8.5 cm612.0 cm) to rear Epicephala larvae. We then recorded rostrata with pollen of B. fruticosa. how and when the larvae left fruits to cocoon and emerge. (TIF) The presence of oviposition scars and eggs on fruits was Table S1 Annual fruit stage of Breynia fruticosa and life history of considered to have a one–to–one correspondence [11]. We Epicephala lativalvaris and E. mirivalvata in Yingge Mountain of randomly examined 23 flowers in Hainan (Yingge Mountain) Hainan, China. and 179 flowers in Fujian (Tianzhu Mountain and Wanshi (DOC) Botanical Garden) to search for eggs (Table S3); and randomly examined oviposition scars on 1162 fruits in Hainan (Yingge

PLoS ONE | www.plosone.org 8 July 2012 | Volume 7 | Issue 7 | e41657 Coevolution between Two Epicephala and Two Breynia

Table S2 Annual fruit stage of Breynia fruticosa and Epicephala Table S7 Developed set of Breynia female flowers in hand- rostrata, and life history of E. lativalvaris and E. mirivalvata in Tianzhu pollination experiment. Mountain and Wanshi Botanical Garden of Xiamen, China. (DOC) (DOC) Table S3 Statistics of oviposition scars made by Epicephala Acknowledgments lativalvaris on Breynia fruits at five locations. We appreciate Fuchen Shi for identifying Breynia plants, to Yanru Yin for (DOC) her assistance in the field. Special thanks are given to members of Tianzhu Table S4 Statistics of eggs laid by Epicephala mirivalvata on Breynia Mountain National Forest Park and Yingge Mountain Nature Reserves for female flowers at three locations. their support and kind help in the field. This research was supported by the National Natural Science Foundation of China (No. 30930014). (DOC) Table S5 GenBank accession number of Epicephala moths COI Author Contributions sequences. Conceived and designed the experiments: HL SW. Performed the (DOC) experiments: JZ BH XY ZW. Analyzed the data: JZ HL. Wrote the Table S6 GenBank accession number of Phyllantheae plants paper: JZ SW HL. matK sequences. (DOC)

References 1. Janzen DH (1979) How many babies do figs pay for babies? Biotropica 11: 48– 14. Janzen DH (1980) When is it coevolution. Evolution 34(3): 611–612. 50. 15. Machado CA, Robbins N, Gilbert MTP, Herre EA (2005) Critical review of host 2. Weiblen GD (2002) How to be a . Annual Review of Entomology 47: specificity and its coevolutionary implications in the fig/fig–wasp mutualism. 299–230. Proceedings of the National Academy of Sciences USA (suppl. 1): 6558–6565. 3. Kato M, Takimura A, Kawakita A (2003) An obligate pollination mutualism and 16. Li BT, Gilbert MG (2008) Flora of China. Missouri Botanical Garden Press. reciprocal diversification in the tree Glochidion (Euphorbiaceae). Proceed- USA 11: 635 pp. ings of the National Academy of Science 100 (9): 5264–5267. 17. Zhang J, Hu BB, Wang SX, Li HH (2012) Six new species of the genus Epicephala 4. Pellmyr O (2003) Yuccas, yucca moths, and coevolution: a review. Annals of the Meyrick, 1880 (: Gracillariidae) associated with Euphorbiaceae Missouri Botanical Garden 90: 35–55. plants. Zootaxa 3275: 43–54. 5. Ehrlich PR, Raven PH (1964) Butterflies and plants: a study in coevolution. 18. Corlett RT (1993) Reproductive phenology of Hongkong shrubland. Journal of Evolution 18: 586–608. Tropical Ecology 9(4): 5012510. 6. Wiebes JT (1979) Co-evolution of figs and their pollinators. Annual 19. Doyle JJ, Doyle L (1987) A rapid DNA isolation procedure for small amounts of Review Ecology and Systematics 10: 1–12. fresh leaf tissue. Phytochemical Bulletin 19: 11215. 7. Pellmyr O, Leebens-Mack J, Huth CJ (1996) Non-mutualistic yucca moths and 20. Park DS, Suh S-J, Oh HW, Hebert PDN (2010) Recovery of the mitochondrial their evolutionary consequences. Nature 380: 155–156. COI barcode region in diverse Hexapoda through tRNA-based primes. BMC 8. Kawakita A, Kato M (2006) Assessment of the diversity and species specificity of genomics 11: 423. the mutualistic association between Epicephala moths and Glochidion trees. Molecular Ecology 15: 3567–3582. 21. Samuel R, Kathriarachchi H, Hoffmann P, Barfuss MH, Wurdack KJ, et al. 9. Kawakita A, Kato M (2009) Repeated independent evolution of obligate (2005) Molecular phylogenetics of Phyllanthaceae: evidence from plastid MATK pollination mutualism in the Phyllantheae2Epicephala association. Proceedings of and nuclear PHYC sequences. American Journal of Botany 92(1): 1322141. the Royal society B-Biological Sciences 276 (1656): 4172426. 22. Brower AVZ (1994) Rapid morphological radiation and convergence among 10. Kawakita A (2010) Evolution of obligate pollination mutualism in the tribe races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA Phyllantheae (Phyllanthaceae). Plant Species Biology 25: 3219. evolution. Proceedings of the National Academy of Sciences 91: 649126495. 11. Kawakita A, Kato M (2004) Obligate pollination mutualism in Breynia 23. Caterino MS, Sperling FA (1999) Papilio phylogeny based on mitochondrial (Phyllanthaceae): further documentation of pollination mutualism involving cytochrome I and II genes. Molecular Phylogenetics and Evolution 11(1): Epicephala moth (Gracillariidae). American Journal of Botany 91(9): 1319–1325. 1222137. 12. Okamoto T, Kawakita A, Kato M (2007) Interspecific variation of floral scent 24. Kathriarachchi H, Samuel R, Hoffmann P, Mlinarec J, Wurdack KJ, et al. composition in Glochidion and its association with host-specific pollinating seed (2006) Phylogenetics of the tribe Phyllantheae (Phyllanthaceae; Euphoriaceae parasite (Epicephala). Journal of Chemical Ecology 33: 106521081. sensu lato) based on nrITS and plastid matK DNA sequence data. American 13. Svensson GP, Okamoto T, Kawakita A, Goto R, Kato M (2010) Chemical Journal of Botany 93(4): 6372655. ecology of an obligate pollination mutualism: testing the ‘private channel’ 25. Swofford DL (2005) PAUP*: Phylogenetic analysis using parsimony (*and other hypothesis in the Breynia2Epicephala association. New Phytologist 186: methods), version 4.0b10. Sinauer. Sunderland. Massachusetts. USA. 99521004.

PLoS ONE | www.plosone.org 9 July 2012 | Volume 7 | Issue 7 | e41657